Rainfall-Runoff Modeling Using SWAT Hydrologic Model: Case Study of DabusRiver Watershed, Ethiopia

Abstract

Rainfall-Runoff modelling is essential for effective and sustainable water resources planning and management of the watershed. The derivation of relationships between the rainfall over a catchments area and the resulting flow in a river is fundamental problem for the hydrologist. In most developing countries, like Ethiopia there are usually no plenty of rainfall records. Therefore objective of this study was to achieve the runoff simulation and investigate rainfall runoff relationship in the study area. This study was carried out on Dabus watershed which is located in the western Ethiopia. Weather data of six stations for a period of 1994 to 2016 were collected and combined with other maps of the study area, such as 30m x 30m Digital Elevation Model (DEM), land use, soil and slope as input data for Soil and Water Assessment Tools (SWAT) model which works in conjunction with Arc GIS. In this study the watershed area was delineated using the SWAT model and then divided into 75 sub-basins. Land use land cover, soil and slope were overlaid to the delineated watershed, and then these sub basins are further divided into 323 HRUs which stands for Hydrological Response Unit. Then by using 23 years of daily weather data SWAT simulation was done for monthly basis to find out Runoff volume and runoff for corresponding Rainfall. After running the model, the Sequential Uncertainty Fitting (SUFI-2) algorithm in SWAT calibration and uncertainty program (SWAT-CUP) were used to evaluate the data uncertainty and for model calibration and validation by using (1996- 2016) years observed stream flow at Dabusnear@Asossa gauging station. The first two years (1994-1996) for warm-up and the next (1996-2008) for the calibration and finally (2009-2016) were used for the validation period. The simulated average annual surface runoff was 228.74 mm. The model performance evaluation statistics showed that during calibration, monthly results were 0.88 and 0.80 for coefficient of determination (R2 ) and Nash–Sutcliffe model efficiency (NSE) respectively. During validation monthly results were 0.88 and 0.76 for coefficient of determination (R2 ), and Nash–Sutcliffe model efficiency (NSE) respectively. The coefficient of correlation (r) for rainfall in a period and the corresponding runoff is found to be 0.9 Hence, it can be concluded that SWAT is able to fairly explain the hydrological characteristics of the Dabus catchment